Visible Yet Unreadable: A Systematic Blind Spot of Vision Language Models Across Writing Systems
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912742079725568 |
|---|---|
| author | Zhang, Jie Xu, Ting Deng, Gelei Hu, Runyi Qiu, Han Zhang, Tianwei Guo, Qing Tsang, Ivor |
| author_facet | Zhang, Jie Xu, Ting Deng, Gelei Hu, Runyi Qiu, Han Zhang, Tianwei Guo, Qing Tsang, Ivor |
| contents | Writing is a universal cultural technology that reuses vision for symbolic communication. Humans display striking resilience: we readily recognize words even when characters are fragmented, fused, or partially occluded. This paper investigates whether advanced vision language models (VLMs) share this resilience. We construct two psychophysics inspired benchmarks across distinct writing systems, Chinese logographs and English alphabetic words, by splicing, recombining, and overlaying glyphs to yield ''visible but unreadable'' stimuli for models while remaining legible to humans. Despite strong performance on clean text, contemporary VLMs show a severe drop under these perturbations, frequently producing unrelated or incoherent outputs. The pattern suggests a structural limitation: models heavily leverage generic visual invariances but under rely on compositional priors needed for robust literacy. We release stimuli generation code, prompts, and evaluation protocols to facilitate transparent replication and follow up work. Our findings motivate architectures and training strategies that encode symbol segmentation, composition, and binding across scripts, and they delineate concrete challenges for deploying multimodal systems in education, accessibility, cultural heritage, and security. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_06996 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Visible Yet Unreadable: A Systematic Blind Spot of Vision Language Models Across Writing Systems Zhang, Jie Xu, Ting Deng, Gelei Hu, Runyi Qiu, Han Zhang, Tianwei Guo, Qing Tsang, Ivor Computer Vision and Pattern Recognition Artificial Intelligence Writing is a universal cultural technology that reuses vision for symbolic communication. Humans display striking resilience: we readily recognize words even when characters are fragmented, fused, or partially occluded. This paper investigates whether advanced vision language models (VLMs) share this resilience. We construct two psychophysics inspired benchmarks across distinct writing systems, Chinese logographs and English alphabetic words, by splicing, recombining, and overlaying glyphs to yield ''visible but unreadable'' stimuli for models while remaining legible to humans. Despite strong performance on clean text, contemporary VLMs show a severe drop under these perturbations, frequently producing unrelated or incoherent outputs. The pattern suggests a structural limitation: models heavily leverage generic visual invariances but under rely on compositional priors needed for robust literacy. We release stimuli generation code, prompts, and evaluation protocols to facilitate transparent replication and follow up work. Our findings motivate architectures and training strategies that encode symbol segmentation, composition, and binding across scripts, and they delineate concrete challenges for deploying multimodal systems in education, accessibility, cultural heritage, and security. |
| title | Visible Yet Unreadable: A Systematic Blind Spot of Vision Language Models Across Writing Systems |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2509.06996 |